Re: GCC 4.0 vs. 3.5

[Robert Dewar <dewar at gnat dot com>]

> So, if we're switching to 4.0 it should be because the code we generate
> is far better.  Not that some code is slightly better, or that we have
> the infrastructure we need to generate far better code -- but that much
> actual code is actually far better.

I disagree with this position. Yes, there are some users for whom
performance is critical, but for most users this is secondary to
reliability and stability (that is, a fast unstable version is
not an attractive substitute).

One purpose of a major version number is indeed to alert people to
new features, which can include better code, but is certainly not
limited to this particular improvement.

The second purpose is to alert users to a major change of technology
that may impact stability. In this case, it is that second issue that
is most relevant.